Lesson 1: Introduction to Graph Theory – Arnaud Casteigts (LaBRI/CNRS, France)
In this class, I will review the basics of graph theory, starting from scratch. In the beginning of the class, the scope will be general, and I will present some of the famous problems in graph theory. Then we will focus on problems more closely related to communication networks (e.g. shortest paths, spanning trees, unit disc graphs, and covering problems). The end of the class will open the topic towards the area of highly-dynamic networks (and graphs), which is the topic of the subsequent class.
Lesson 2: Network optimization – Francesca Guerriero (Unical, Italy)
This class will give an overview of the most interesting mathematical models and solution approaches developed in the network optimization field. The attention will be focused on the models and methods, used to address problems arising in the optimal planning and management of telecommunication networks. It will review the main ingredients of the network optimization formulations and the main elements of the related solution approaches.
Topics:
- Representation of a network and its behaviour using mathematical programming tools
- Formulation of optimization problems on network (design, routing, protection)
- General-purpose solvers and ad-hoc algorithms to solve optimization problems on networks
Lesson 3: Large scale experimental platforms – Serge Fdida (LIP6/CNRS, France)
Experimentally driven research has developed in order to equip the community with instruments that can assist them with testing their design and solutions. Testbed as a Service appeared to bridge this gap and propose now matured solutions for that purpose. OneLab/FIT, GENI and Fed4Fire+, are examples of such solutions that have hosted thousands of experiments and hundreds of users. They are now extended and enriched through new initiatives such as SLICES, NSF/PAWR, CENI and ICT-17 projects.
This talk will emphasize the potential for using these platforms with novel objectives. Instead of providing access to raw resources, the facility can accelerate the testing of devices for specific concerns. The talk will illustrate this claim with examples taken from the F-Interop and Armour projects focusing on IoT.
Lesson 4: Queuing Systems – Alain Jean-Marie (Inria, France)
The class will review the principles of modelling communication systems with queues, and the methods of analysis that are associated. IT will review the main elements of queueing systems such as: Arrival processes, service, scheduling, blocking etc. It will focus on their modeling with discrete-time Markov chains and discuss the different methods for obtaining quantitative information from the models.
Topics:
- Basic features of queueing systems
- Metrics associated with queuing-modeling queueing systems with Markov chains
- Numerical solution and simulation
Lesson 5: Markov decision process – Emmanuel Hyon (LIP6/CNRS)
The class will give an overview of the problems related to the optimization and the dynamic control of stochastic systems (and especially queueing system). This class will present the Markov decision Process Framework and its use for discrete time problems. The main elements of such processes (state space, action space, transition probabilities, decision rule) will be detailed as well as their classification related to their optimization criteria. It will also present different algorithms to solve a CDM and to find numerically the optimal decision rule. Topics:
- Dynamic control of queues
- Modeling with Markov decision process
- Solving Markov decision processes
